The worst good idea ever

OK, maybe this isn't actually the worst, but this is a really bad idea. A couple of Congressmen are proposing an amendment to an energy bill that would extend daylight-saving time by 2 months. The idea is that more people are awake between 5 and 6 PM than between 5 and 6 AM. So, if you have daylight go later for longer, then more people could go without turning on the lights in the early evening. I.e., you would reduce the energy consumption for early evening light.

While I applaud the idea of using radical means to reduce energy consumption, this particular suggestion is a very bad idea and I hope it fails. Messing with daylight-saving time is not just a simple matter of remembering to set your clock back in November instead of September. Instead, this would involve updating the operating system on every computer in the world. The alternative would be to manually change the time zone on your computer from September to November. So I, living in the Eastern time zone in the U.S., would have to set my computer to the Canadian Atlantic time zone for those two months, before switching back to Eastern.

As a means of comparison, consider Y2K. The problem was that programmers got the time from the operating system, then stored the year as a 2 digit number instead of a 4 digit one. But the operating system was set to have the right time. Not all programs did this, though. The billions spent on Y2K involved looking at the programs, determining if it was affected, and fixing it if so. With this daylight-saving change, if there is a single program on your computer affected, then all of your programs are affected. The reason is that you have to update the operating system for that one program, which in turn means a new version of the operating system for all the others. This is because operating system updates do not come as small, stand-alone pieces. Instead, you get things like a Microsoft XP Service Pack. That SP will contain the updated clock, but it will also contain thousands of other updates, any of which could cause your computer or program to crash. On my team, when we updated the operating system we were using (going from AIX 4.3.3 to AIX 5.1), we spent 6-8 months doing qualification testing.

So how much of a problem is having the wrong time on a computer? Imagine showing up for a flight only to find that it left an hour ago because the airport's computers had the wrong time. Imagine an engineer setting the wrong trajectory on a new satellite, causing it to crash into an existing one, because his computer had the wrong time setting. There are many scenarios where things like this could happen. Sure, your first reaction is to think that the engineer mentioned here would be smart enough to avoid a mistake like that. That is, until you remember that brilliant (and, yes, I mean that sincerely and without sarcasm) engineers at NASA lost data on a Huygens experiment because they forgot to turn a single instrument on. Other scientists have caused huge failures because they calculated values using English measurements instead of metric ones.

So, unless they create the goal of implementing this change about 20 years in the future, I think this is a really bad idea and needs to fail. There are other, more feasible ways to save energy. How about tax credits for buying a car with over 30 mpg and tax debits for buying one with less than 20 mpg? Make the owner of that Hummer pay an extra $30K per year in taxes. That'll certainly make people think twice before buying a fuel inefficient vehicle. As a side note, yes, there are tax credits for buying an electric or hybrid car. However, there are non-electrics that also get incredibly good gas mileage, some nearly as good as hybrids. The reward should be for reducing oil usage, not exclusively for supporting an emerging technology.